Вы находитесь на странице: 1из 6

Aircraft Engineering and Aerospace Technology

Criteria for self loosening of fasteners under vibration


Article information:
To cite this document:
GERHARD H. JUNKER, (1973),"Criteria for self loosening of fasteners under vibration", Aircraft Engineering and Aerospace Technology, Vol. 45 Iss 1 pp. 12 - 17
Permanent link to this document:
Downloaded on: 01 January 2015, At: 10:14 (PT)
References: this document contains references to 0 other documents.
To copy this document: permissions@emeraldinsight.com
The fulltext of this document has been downloaded 76 times since 2006*

Users who downloaded this article also downloaded:

Muttukrishna Sarvananthan, (2008),"Economic imperative for peace in Sri Lanka", Contributions to Conflict Management, Peace Economics and Development, Vol. 5 pp. 171-184
GERHARD H. JUNKER, (1972),"Criteria for self loosening of fasteners under vibration", Aircraft Engineering and Aerospace Technology, Vol. 44 Iss 10 pp. 14-16 http://

Access to this document was granted through an Emerald subscription provided by 281668 []

For Authors
If you would like to write for this, or any other Emerald publication, then please use our Emerald for Authors service information about how to choose which publication to write
for and submission guidelines are available for all. Please visit www.emeraldinsight.com/authors for more information.

About Emerald www.emeraldinsight.com

Emerald is a global publisher linking research and practice to the benefit of society. The company manages a portfolio of more than 290 journals and over 2,350 books and book
series volumes, as well as providing an extensive range of online products and additional customer resources and services.

Downloaded by Purdue University At 10:14 01 January 2015 (PT)

Emerald is both COUNTER 4 and TRANSFER compliant. The organization is a partner of the Committee on Publication Ethics (COPE) and also works with Portico and the
LOCKSS initiative for digital archive preservation.
*Related content and download information correct at time of download.




January 1973

Criteria for self loosening of

fasteners under vibration

Downloaded by Purdue University At 10:14 01 January 2015 (PT)


E-F-N curves
Symbols and abbreviations
= Displacement of clamped components under transverse
force, mm or in.
= Displacement of U-shaped top part of vibration machine
in unloaded condition (zero load amplitude), mm or in.
= Major diameter of thread, mm or in.
= Pitch diameter of thread, mm or in.
= Dilation of nut under axial load, mm
= Amount of embedding (Brinelling, 'Setzen', plastic de
formations) during dynamic dynamic loading of a
joint, mm or in.
= Tightening or torque factor, which considers the preload
deviations caused by differences of friction coefficients
and applied torque in the design equation for bolted
= Spring constant of load transmitting parts of vibration
machine, fig 16(a); kgf/mm or lb/in.
= Spring constant of clamped plates or components, kgf/
mm or lb/in.
= Spring constant of screw, kgf/mm or lb/in.
= Friction diameter of a bolt head or nut, mm or in.
= Energy for elastic deflection of a bolt in a transversely
loaded joint during cycle, mm kgf or in./lb.
= Energy absorbed by friction in a sliding joint during
cycle, mm kgf or in./lb
E max
== Maximum output of vibration machine for a certain
eccentric adjustment, mm kgf or in./lb.
= Axial force in a fastener, kgf or lb
Maximum axial force in a prestressed fastener under
working load, kgf or lb
= Residual clamping force necessary for maintaining grip
friction of clamped components, kgf or lb
= Transverse force in a bolted joint as well as in the vibra
tion machine, kgf or lb
= Transverse force in vibration machine with eccentric
adjustment d0, kgf or lb
= Preload in a fastener, kgf or lb
= Working load in a bolted joint, kgf or lb
= Loss of preload in a bolted joint by embedding, kgf or lb
= Weight, kgf or lb
= Transverse for moving loads resting on a horizontal
plane and a slope, kgf or lb
Off-torque for a threaded fastener under static conditions,
m kgf or in./lb
= Vibration product equals displacement 'd' times trans
verse force F T in a transversely vibrated joint, a
significant factor for the locking performance of a
self-locking fastener, mm kgf or in./lb
= Angle of load deflection curve of load transmitting
machine parts, deg

= Slope angle and helix angle, deg

= Coefficient of friction in threads, = t a n p

= Coefficient of friction in threads, under head or nut
E-F-N curves = Curves of constant preload for certain vibration energy
levels versus number of cycles
F T - N curves = Transverse force versus number of cycles, significant for
surface integrity of clamped components
Fy-N curves = Decreasing preload under transverse vibration versus
number of cycles
V-N curves = Vibration product versus number of cycles

This is the continuation of the article by Gerhard H. Junker, of the European

Research and Engineering Standard Pressed Steel Co, Unbrako. The first part,
which appeared in October, covered the mechanism of self loosening. The E-F-N
curves, surface integrity and test programme will be covered in the final part of
this feature.
CCORDING to previous research work, it has to be assumed that
total self-loosening docs not occur in joints that arc dynamically
loaded only in an axial direction. This may be at least valid in all
cases where the joint is properly tightened. But it has been proven that at
least partial rotation of the nut can occur when the ratio between working
load and preload is high, and particularly if the load in the thread is largely
diminished by the compression amplitude of an alternating working load.
The partial loosening by rotation leads to a further loss of preload, and
thus to a stop of the rotating process, because the internal off-torque is
proportional to the preload, which itself is reduced by the rotation process.
On the other hand, the loss of the preload increases the danger of fatigue
failure because the total amount of alternating working load (axial) is felt
by the bolt.
On the other hand, total rotational loosening can occur in transversely
loaded connections as soon as the clamping load can no longer maintain
sufficient grip friction so that transverse slippage between the clamped
parts occurs. Transverse slippage, as a matter of fact, is quite common
because it is caused not only by transverse loads or transverse components
of a load in any direction, but also by coaxial loads that deform the
clamped part elastically (that is, by all joints that are loaded eccentrically,
such as bearing caps). The proposed concept of the mechanism of selfIooscning or self-rotating of fasteners completes the picture of bolt
failures under vibration by introducing the 'self-rotating process'. Many
cases of fatigue failures under obviously very low preload, which could not
be explained by plastic deformations or even by insufficient tightening,
were probably caused by partial self-rotating.
The failure mechanism of dynamically loaded bolted connections can
have 12 different lines, which lead cither to fatigue failure of the bolt
(infrequently the clamped part or the nut) or to self-loosening of the
fastener. This shows that the basic rule for avoiding bolt failure, either by
fatigue breakage or by self-loosening, is to supply and to maintain suffi
cient clamping load. Since this article deals with the 'self-loosening'
branch of the failure mechanism, the design rule for avoiding self-loosen
ing or loss of preload by partial self-rotating is as follows: Design in such
a way that no relative movement occurs between the contact bearing
surfaces normal to the axis of the bolt or between the thread flanks of the
fastener components.
This requires that a sufficient residual clamping force remains when the
scatter of preload values after tightening and the embedding phenomena
(Setzen), due to plastic deformations of the fastener or the clamped parts,
is taken into consideration. The residual clamping load should never fall
short of the value F T / (transverse load/friction coefficient) to make sure
that no transverse slippage can occur. Modern rules for calculation of
bolted connections 2/3 are therefore based on the necessary residual
clamping force, the estimated degree of embedding (Setzen), and the
scatter of the preload obtained by tightening, as represented in the following

Tabulated values for the different factors are given.

As it was previously shown, using the example of the crown wheel,
there exist cases where these demands can be met only if the joint fasteners
arc uneconomically overdimensioned. There is no doubt that many
dynamically loaded bolted connections arc beyond an exact calculation;

January 1973




Downloaded by Purdue University At 10:14 01 January 2015 (PT)

therefore the relative movements cannot always be excluded by means of

the joint design only. In these cases, use is recommended of fasteners with
self-locking elements that tend to resist the internal off-torque which
occurs when inherent self-locking is lost.
Proposed test methods and machinery
Since relative movements between clamped parts and fasteners cannot
be avoided by means of the joint design only, and self-locking fasteners
have to be used for preventing 'rotating loose', a criterion has to be found
for the effectiveness of such self-locking fasteners. Several test methods
have been tried in the past and are still in use4"3. Most of them are more
or less qualitative methods, which result in 'rotating loose' or 'not rotating
loose' under certain specified conditions (some of them in preloaded
condition; some in an unpreloaded condition). It is obvious that the
latter ones are practical only for prevailing torque types of locking
fasteners. Other specifications give values for installation torque, break
away torque, and prevailing torque.
With this paper an entirely different approach is proposed. Since
transversely loaded joints tend more to self-loosening, the test procedure
suggested imitates these actual conditions. The first attempts were made
with a device consisting of two parts clamped together by the specimen,

with load cells and a displacement pickup to record transverse load,

preload, and displacement.
The product, transverse force times displacement, was called vibration
energy. Its maximum values were significantly different for the various
locking elements.
This test method yielded results that could not be reproduced, since the
machine used for the tests was a resonant type of fatigue machine. This
type generates a force that is a function of the machine frequency. When
starting a test, the transverse force increased with increasing frequency
as a function of time until the loosening process of the tested fastener
was initiated. Depending upon the starting speed of the machine, a
different maximum force (and therefore maximum energy) was necessary
for rotation loosening of a nut or bolt. Therefore, a new vibration machine
was designed for these tests.
The prototype of the new machine was built by the Corporate Machine

Building Dept. of the SPS Co. This machine generates a transverse

sliding motion between two clamped parts by means of an adjustable
eccentric. The resulting transverse force is independent of frequency and
starting speed of the machine. It is for a given machine stiffness (spring
constant of the load transmitting parts). strictly a function of the eccen
tric adjustment.
Fig 1 shows the heart of the machine. The tested bolt clamps the
U-shaped top part onto the bed, an integral part of the machine frame.
The clamping force is measured with a compression load cell through
which a bushing with internal threads is placed for testing bolts. When
testing nuts, this bushing may hold a threaded stud, or a special bushing
with external threads may be used.
The U-shaped top part is displaced parallel to the bed. The force is
transmitted by a load cell mounted on flexure plates, a connecting rod
being attached to the load cell by crosswise flexure plates and by an adjust-




January 1973

able eccentric. The U-shaped top part is separated from the bed by flat
strips of needle bearings, to avoid galling.

the machine housing. A V-belt drive and different pulleys allow testing
frequencies of 1500, 3000, 4500cpm.

The relative movement between bed and top part is measured with
special linear differential transformers. The tightening and loosening
angles were measured with a linear potentiometer, which was attached to
the test specimen by a flexible shaft, Exchangeable threaded bushings
and inserts in the U-shaped top part make the mechanism usable for
testing bolts from in (or M6) up to in (or M16) thread sizes. This
range covers the majority of the common locking elements.

Fig 2 shows three different applications for the machine. In the upper
picture the U-shaped top part is connected to the load cell with a bolted
plate, to eliminate all play. The arrangement is used for tests with pure
transverse loading. This plate is replaced in the middle picture by an
U-shaped sheet metal specimen that transmits the transverse forces when
clamped with the U-shaped top part to the bed. With this arrangement,
the fatigue strength of bolted sheet metal joints can be evaluated.
This is of special interest if the fastener damages the surface of
the clamped parts, for example, when using locking fasteners with ser
rated bearing surface. When the connecting plate is replaced by a con
necting rod, it is possible to rotate the bed so that it forms any
angle between 90 and 180deg with the frame of the machine. The centre
of rotation coincides with one bearing of the second connecting rod. This
arrangement allows loading the test specimens by any combination of
transverse and axial forces.

The machine is driven by a 36kW (5hp) ac motor, which is located in

The recording equipment used for the tests described above is shown
in fig 3. Transverse force F T and displacement 'd' are plotted on a Visicorder as a function of time. The bolt preload is recorded on a third
channel. By using a two channel oscilloscope, the transverse force FT is
recorded as a function of displacement 'd' in X-Y hookup on one channel
and the preload on the second channel. Running the tests with a special
low speed drive (10cpm), the same information was plotted on an X-Y1-Y2
The electrical values of transverse force FT and displacement 'd' were
multiplied by a specially built electronic multiplication circuit. The
result F T d is plotted as a function of preload on an X-Y1-Y2 recorder.

Fig 4
Dimensional Specification of Tested Specimens


Downloaded by Purdue University At 10:14 01 January 2015 (PT)

A Nonlocking Screws
Hex head cap screws,
DIN 933, M10 x 15 x 30


Socket head cap screws,

1960 Series,
-16 x 1-in


Socket head cap screws,

1960 Series,
-16 x 1-in


Pitch: l5mm
Thread tol: Bolt 6g
Nut 6H
Nominal max. thread
allowance: 0344mm
Pitch: 1586mm
Thread tol: Bolt 2A
Nut 2B
Nominal max. thread
allowance: 0290mm
Pitch: 1586mm
Thread tol: Bolt 3A
Nut 3B
Nominal max. thread
allowance: 0193mm

B Spring Washers
Hex head cap screws,
DIN 933 (A1) plus
spring washer DIN 128
C Free Spinning Locking Screws
Flange head, hex cap screws,
M10 x 15 x 30, with
triangular teeth, groove
and bearing platform
near the shank
Flange head hex cap screws,
M10 x 15 x 30, with long
radial teeth and
circumferential bearing
ring, edges of teeth and
bearing ring forming one
continuous curve

Thread tol: Bolt 6g

Nut 6H
Nominal max. thread
allowance: 0344mm
Thread tol: Bolt 6g
Nut 6H
Nominal max. thread
allowance: 0344mm



The tightening and loosening angle measured with a linear potentiometer

is recorded as a function of preload on the second channel of the X-Y1-Y2
To check out the machine and to evaluate the possibilities and limita
tions of the various testing, measuring, and recording methods, the test
specimens were restricted to nonlocking screws of different thread toler
ance (with and without spring washers) and to two types of free spinning


January 1973

bearings between the clamped plates merely prevents galling; it does not
favour either group of locking elements.
The diagrams recorded during tests have basically the same shape as
the schematic drawing of fig 6. Fig 7 shows the transverse force-displace
ment diagram of a non-locking screw A1 and of a free spinning locking
screw C2, recorded at a machine speed of 10cpm. The diagram is presented
for 100, 80, 60, 20 and 0 per cent of the original bolt preload, which is also

Fig 5
Mechanical properties of specimen
(Diamond pyr
300 load)

Downloaded by Purdue University At 10:14 01 January 2015 (PT)


Hardness Hardness














Mech. properties of
screw same as A1
Case hardness mea
sured 005mm
(000197in.) from

self-locking screws with serrated bearing surfaces. This restriction has

been made because of lack of time and because these types of screws are
used in great numbers for sheet metal constructions, particularly in the
automotive industry. The described test methods and machinery can be
used for testing all kinds of self-locking nuts and also self-locking screws
of the prevailing torque and adhesive type. A comprehensive study with
specimens of all groups of self-locking fasteners is in preparation. Geo
metric shapes and dimensions of the test specimens are given in fig 4 and
the mechanical properties in fig 5. Screws of the size M10 x 15 and -16
have almost equal nominal thread diameter, pitch diameter, and pitch so
that their test results can be compared.
All fasteners were oiled before testing and were used with plain washers
under the head, having a Rockwell hardness of 77Rb (HB30= 137kgf/mm2).
The free spinning types of locking fasteners were additionally tested on
washers of higher hardness (Rc=39-41).
During all tests the specimens were originally preloaded to 2500kgf=
55001b. This relates to 75 per cent of the proof load of screws with a
strength level of ISO Class 8.8 (similar to Grade 5). All tests except those
run with a special low speed drive (10cpm) were run with a frequency of
X-Y plot of transverse force and displacement
It can be seen that the distance W1 for threads with close tolerance is
smaller than the distance W2 for threads with wide tolerance. Therefore,
screws with close thread fit must have a higher resistance to self-rotating
loosening than screws with wide thread fit. These considerations explain
that the prevailing torque type of locking screw works by reducing or
eliminating the thread allowance by various means. The self-forming
screws must be added to this category for the same reason.
According to fig 6 a bolted connection that slips under the bolt head
will absorb the most energy; the hysteresis loop encloses a considerable
larger area. Where the clamped parts arc not separated by flat strips of
needle bearings and where they arc clamped together directly, the inter
action of force, deflection, and displacement is basically the same. Merely
the friction energy between the clamped parts (Fv2dmax) has to be gen
erated additionally every cycle. When the transverse force F T is large
enough to force a relative displacement of the clamped plates. the same
process as that described in fig 6 takes place. The use of strips of needle

shown on the same plot as a function of displacement (upper part of

chart). It can be seen that the nonlocking screw A1 starts to slip under
the head in the fully preloaded condition; however, the screw C2 with
serrated head bearing surface does not start to slip until the preload is
decreased to 40 per cent of the original preload. The tangent curves in
fig 7 represent the load/deflection curve (spring constant CM) of the load
transmitting parts of the machine.
Tests show that the self-loosening of screws is independent of frequency.
It simply depends upon the occurrence of relative motions and on the

January 1973




Downloaded by Purdue University At 10:14 01 January 2015 (PT)

length of such motions during one cycle. If the forces causing these relative
motions are inertia forces, they are a function of the square of the fre
quency. In this case the frequency indirectly influences the loosening
Tests yielded interesting results when recording the angle of rotation
during loosening in addition to the force/displacement hysteresis curve.
It becomes apparent that the loosening process of a nonlocking fastener
starts with the very first cycle. The recorded angle of rotation loosening
was 9deg after 25 cycles; it was constant, since 036deg rotation occurred
per cycle. The recording of preload and angle of rotation as a function of
displacement during 10cpm tests gave an answer to the question of what
causes the relative high loss of preload of all free spinning, self-locking
fasteners at the beginning of the vibration process. This phenomenon
was especially encountered at test series described later.
Fig 8 shows the first 50 cycles at 10cpm frequency for free spinning
locking screws C1 and C2 (the first five cycles are recorded, then every
fifth; screws seated on a soft washer). The screws lose 25 per cent and
144 per cent respectively of their original preload during the test. Re
cording the angle of rotation showed that the high loss of preload at the

so that X-Y recorders may be employed. It is not so well suited for large
test series at high testing frequencies when a statistical evaluation is
desired, since in that case the hysteresis curves will have to be photographed
from an oscilloscope. Fig 9 shows maximum output of the vibration
machine and actual hysteresis curves.
beginning was not only the result of the 'digging in' of the teeth but also
of a rotation of 13deg for C1 and 06deg for C2. After 30 cycles, the rotat
ing process stops and further loss of preload is due to brinelling of the
teeth into the surface.
The loss of preload after 50 cycles is the same when the screws are
seated on hardened washers; however, it is caused to a larger degree by
rotation. That this process is dependent largely on the shape of the teeth
becomes obvious by the following: The screws C2 having long rectangular
teeth lost only half as much preload after 50 cycles under equal conditions
and turned only half the angle as the screw C1, which has triangular high
Fig 8 (C2) also shows clearly the two bends in the hysteresis curve
which characterise the limits of thread slippage. The hysteresis curve
recorded by the X-Y recorder shows a screw that does not slip under the
head. This kind of test is primarily suited for evaluating rotating and
brinelling processes. It can be used especially well at slow testing speeds

1. G. Junker and G. Meyer,'NeuereBetrachtungen ber die Haltbarkeit von dynamisch belasteter
Schraubenverbindungen' (New Aspects on the Fatigue Behaviour of Dynamically Loaded
Bolted Joints), Draht-Welt, Fachbeilnge Schrauben Muttern Formteile, 54 (1968), Nr. 7,487-499
2. G. Junker and D. Blume, 'Neue Wege einer systematischen Schraubenberechnung' (Modern
Rules for Calculation of Bolted Connections), Dusseldorf: Triltsch, 1965.
3. G. Junker, 'Neuc Prinzipien der Schraubenberechnung (New Principles of Calculation for
Bolted Connections), Maschinrnmarkt, 71 (1965), Nr. 74, 16-29.
4. AN-N-10a, 'Nuts; self-locking. 550F'.
5. AN-N-5b, 'Nuts; self-locking, 250F'.
6. MIL-N-787317. 'Nut; self-locking, 1200F'.
7. MS 26531 (ASG), 'Vibration Test RIG'.
8. NAS 3350, 'Nuts; self-locking 450F, High Quality'.
9 MIL-N-25027C, 'Nut; self-locking, 250F, 450F, and 800F, 125 KSI FTU, 60 KSI FTU and
30 KSI F T U ' .
10. MIL-F-18240, 'Self-locking element for fastener externally threaded 250F'.
11. IFI-100, 'Prevailing-Torque Type Steel Hexagon Locknuts'.
12. IFI-101, 'Torque-Tension Requirements for Prevailing Torque Type Steel Hexagon Locknuts'
13. MIL-STD-1312, 'Fasteners, Test Methods'